计算Spark ML堆需求

Question

我试图在我的训练样本上训练一个朴素的bayes模型。对于每一个观察，我都有一个SparseVector，大约有27000个特性。我通过ipython笔记本使用spark2.0。在

我在一个有4个工人的集群上运行它，每一个都有一个带有40GB内存和8个核心的执行器。驱动程序有32GB的内存。Spark UI显示数据集的大小为18.4GB，分布在100个分区中。在

模型拟合开始了，但是过了一会儿，我得到了下面的OOM错误。在

我正试图找出一种方法来估计执行此装配所需的RAM。有可靠的方法来计算估算吗？在

另外，我分配40gb给每个执行器，并可以在spark Master web UI页面上进行确认。但是，当我进入应用程序详细界面并单击Executors时，它显示每个执行器只有21.2GB。为什么会这样？在

提前谢谢

这是垃圾场：

Py4JJavaError: An error occurred while calling o2392.fit.
: java.lang.OutOfMemoryError: Java heap space
    at scala.collection.mutable.ArrayBuilder$ofDouble.mkArray(ArrayBuilder.scala:518)
    at scala.collection.mutable.ArrayBuilder$ofDouble.resize(ArrayBuilder.scala:524)
    at scala.collection.mutable.ArrayBuilder$ofDouble.ensureSize(ArrayBuilder.scala:536)
    at scala.collection.mutable.ArrayBuilder$ofDouble.$plus$plus$eq(ArrayBuilder.scala:549)
    at scala.collection.mutable.ArrayBuilder$ofDouble.$plus$plus$eq(ArrayBuilder.scala:511)
    at scala.collection.mutable.ArrayOps$$anonfun$flatten$2.apply(ArrayOps.scala:83)
    at scala.collection.mutable.ArrayOps$$anonfun$flatten$2.apply(ArrayOps.scala:82)
    at scala.collection.IndexedSeqOptimized$class.foreach(IndexedSeqOptimized.scala:33)
    at scala.collection.mutable.ArrayOps$ofRef.foreach(ArrayOps.scala:186)
    at scala.collection.mutable.ArrayOps$class.flatten(ArrayOps.scala:82)
    at scala.collection.mutable.ArrayOps$ofRef.flatten(ArrayOps.scala:186)
    at org.apache.spark.mllib.classification.NaiveBayesModel.<init>(NaiveBayes.scala:56)
    at org.apache.spark.mllib.classification.NaiveBayes.run(NaiveBayes.scala:433)
    at org.apache.spark.mllib.classification.NaiveBayes$.train(NaiveBayes.scala:507)
    at org.apache.spark.ml.classification.NaiveBayes.train(NaiveBayes.scala:106)
    at org.apache.spark.ml.classification.NaiveBayes.train(NaiveBayes.scala:76)
    at org.apache.spark.ml.Predictor.fit(Predictor.scala:90)
    at org.apache.spark.ml.Predictor.fit(Predictor.scala:71)
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.lang.reflect.Method.invoke(Method.java:498)
    at py4j.reflection.MethodInvoker.invoke(MethodInvoker.java:237)
    at py4j.reflection.ReflectionEngine.invoke(ReflectionEngine.java:357)
    at py4j.Gateway.invoke(Gateway.java:280)
    at py4j.commands.AbstractCommand.invokeMethod(AbstractCommand.java:128)
    at py4j.commands.CallCommand.execute(CallCommand.java:79)
    at py4j.GatewayConnection.run(GatewayConnection.java:211)
    at java.lang.Thread.run(Thread.java:745)